Let y be a transient state. Show that for all x
$$\sum_{n=0}^{\infty}P^n(x,y) \leq \sum_{n=0}^{\infty}P^n(y,y)$$

Hint: Use $\sum_{n=1}^{\infty}P^n(x,y)=\frac{\rho_{xy}}{1-\rho_{yy}}$

I tried to sum one and use the property $0 \leq \rho_{xy} \leq 1$ but I could not conclude. Any ideas?

  • $\begingroup$ Doesn't $\sum_{i=1}^\infty P^n(x,y)$ obviously diverge? $\endgroup$ – Kenta S Mar 25 at 20:20
  • $\begingroup$ mmm why? y is transient so there is no problem, right? $\endgroup$ – Bayesian guy Mar 25 at 21:09
  • $\begingroup$ this doesn't make any sense $\sum_{i=0}^{\infty}P^n(x,y) \leq \sum_{i=0}^{\infty}P^n(y,y)$ the indexing in the sum is over i, but $i$ doesn't show up in there, only $n$ and some y, evidently neither of which are related to $i$ $\endgroup$ – user8675309 Mar 25 at 23:50
  • $\begingroup$ my bad, it has to be n instead of i $\endgroup$ – Bayesian guy Mar 26 at 4:33

I did it... We know that $$P^0(x,y)=P[X_0=y|X_0=x]=\begin{cases} 1 & if x=y\\ 0 & if x \neq y \end{cases}$$ In other hand , let's see that \begin{align*} \sum_{n=0}^{\infty}P^n(y,y)&=P^0(y,y)+\sum_{n=1}^{\infty}P^n(y,y)\\ &=1+\sum_{n=1}^{\infty}P^n(y,y)\\ &=1+\frac{\rho_{yy}}{1-\rho_{yy}} \tag{Hint}\\ &=\frac{1}{1-\rho_{yy}}. \end{align*}

Therefore, \begin{align*} \sum_{n=0}^{\infty}P^n(x,y)&=P^0(x,y)+\sum_{n=1}^{\infty}P^n(x,y)\\ &=0+\sum_{n=1}^{\infty}P^n(x,y)\\ &=\frac{\rho_{xy}}{1-\rho_{yy}} \tag{Hint}\\ &\leq \frac{1}{1-\rho_{yy}}. \tag{$0 \leq \rho_{xy} \leq 1$}\\ &=\sum_{n=0}^{\infty}P^n(y,y). \end{align*}

Hence, $$\sum_{n=0}^{\infty}P^n(x,y) \leq \sum_{n=0}^{\infty}P^n(y,y).$$ Q.E.D.


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.